Development and validation of observing‐system simulation experiments at NASA's Global Modeling and Assimilation Office

Abstract

Office (GMAO) are described. The OSSEs mimic the procedures used to analyze global observations for specifying states of the atmosphere. As simulations, however, OSSEs are not only confined to already existing observations and they provide a perfect description of the true state being analyzed. These two properties of the simulations can be exploited to improve both existing and envisioned observing systems and the algorithms to analyze them. Preliminary to any applications, however, the OSSE framework must be adequately validated.This first version of the simulated observations is drawn from a 13 month simulation of nature produced by the European Center for Medium-Range Weather Forecasts. These observations include simulated errors of both instruments and representativeness. Since the statistics of analysis and forecast errors are partially determined by these observational errors, their appropriate modelling can be crucial for validating the realism of the OSSE. That validation is performed by comparing the statistics of the results of assimilating these simulated observations for one summer month compared with the corresponding statistics obtained from assimilating real observations during the same time of year. The assimilation system is the threedimensional variational analysis (GSI) scheme used at both the National Centers for Environmental Prediction and GMAO. Here, only statistics concerning observation innovations or analysis increments within the troposphere are considered for the validation.In terms of the examined statistics, the OSSE is validated remarkably well, even with some simplifications currently employed. In order to obtain this degree of success, it was necessary to employ horizontally correlated observation errors for both atmospheric motion vectors and some satellite observed radiances. The simulated observations with added observation errors appear suitable for some initial OSSE applications.

Evaluation of the Earth Systems Research Laboratory’s global Observing System Simulation Experiment system

“…Most radiance observation errors are horizontally correlated, with the exception of GOES radiance and OSBUV (ozone) observations. The errors differ from those described in Errico et al (2013) as follows: satellite wind errors including both feature-tracking winds and scatterometer winds are uncorrelated; satellite radiance errors are generated using the GSI error tables and are not refined for individual channels; satellite radiance correlation lengths are not calibrated. The impact of using uncorrelated rather than correlated errors for radiance observations is discussed in detail by Errico et al (2013).…”

“…The radiance observations for AMSU-A, AMSU-B, AIRS, HIRS-2, HIRS-3 and MSU generated by GMAO are described in Errico et al (2013). The brightness temperatures were calculated from the Nature Run fields along vertical profiles using the Community Radiative Transfer Model (CRTM) version 1.2.…”

“…Observation errors were added to the perfect synthetic data to introduce both instrument and representativeness errors using an early version of the methods described in detail by Errico et al (2013). Random errors were generated from a normal distribution with mean of zero and SD specified per observation type.…”

“…Since y o can be considered the sum of the 'perfect' observation and the observation error, the gain acts on both the perfect observation and the observation error. Using a 6-hour forecast from the previous cycle as the background state, it has been found (Errico et al, 2013) that x b is relatively insensitive to changes in y o , and thus x a can be manipulated by changing the observation error. A rapid calibration method was used to perform preliminary calibration and tuning of all conventional data types.…”

Improving Flood Prediction Assimilating Uncertain Crowdsourced Data into Hydrological and Hydraulic Models

“…A similar approach, termed "observing system simulation experiment" (OSSE), is commonly used in meteorology to estimate synthetic "true" states and measurements by introducing random errors in the state and measurement equations (Arnold and Dey, 1986;Errico et al, 2013;Errico and Privé, 2014).…”